Problems

Clique covering

1. The results of find clique may overlap with each other

   solution: using package "networkx" to find the max clique , remove the node and so on.

2. After find all clique, each clique represents a core set, some clique only has only element ( 1 trajectory), should we identify it as a core set?

ex.[[3, 2, 10, 11, 0, 9, 6, 12, 13], [1, 4], [5, 7], [8], [16], [14], [15]]

This result may affects the calculation of benefit.

$IncCOOH(K_{m},K_{n}) = COOH(K_{m}) + COOH(K_{n}) + \sum _{c_{i}\epsilon K_{m}} min_{ c_{j}\epsilon K_{n}} \left \{ I(C_{i},C_{j}) \lambda \right \}$

$I(C_{i},C_{j})$ denotes the mumber of vertices that have no edge from core set Ci to Cj and $\lambda$ is the threshold used in the clure-graph.

assum that:

$K_{m}$ is C$_{4}$ = {4}

$K_{n}$ is C$_{1}$ = {3,2,10,11,0,9,6,12,13}

the result of $I(C_{8},C_{1})$ may equals to zero, and C4 seems merge to other cluster easily.